Introduction

Cardiac arrest (CA) is one of the leading causes of death in the world due to its high morbidity and mortality, and conventional cardiopulmonary resuscitation (CCPR) is the primary method for treating patients in CA [1]. However, during CCPR, even with optimal manual chest compressions, the cardiac output is approximately 20% to 30% of the normal value [2]. Moreover, the duration of cardiopulmonary resuscitation (CPR) affected the prognosis of CA patients markedly. As the duration of CCPR prolongs, the survival chances of CA patients gradually become remote. In general, when CCPR lasts longer than 35 min, only a very small number of patients achieve the return of spontaneous circulation (ROSC) and good neurological prognosis ultimately [3]. Overall, most CA patients had a low survival rate and unfavorable neurological prognosis, much less refractory CA, which urges us to find an alternative resuscitation method other than CCPR.

Extracorporeal cardiopulmonary resuscitation (ECPR), also known as extracorporeal membrane oxygenation, is an improved cardiopulmonary bypass supporting heart and lung function. Increasingly, ECPR has been used in the past few years due to the improvement and miniaturization of the technology [4]. ECPR can provide sufficient blood perfusion to vital organs such as the brain and heart during CA until the cardiac output is fully restored, thus avoiding organ failure and reducing the risk of repeated CA caused by ischemia-induced myocardial dysfunction [5]. In addition, using ECPR can extend the window for resuscitation, thus providing more treatment time for medical practitioners [6]. Multiple studies have also suggested that ECPR is a feasible alternative to CCPR, it can provide better prognosis for patients with CA in comparison with CCPR [7,8,9,10,11,12,13,14].

However, ECPR comes with certain limitations. Although ECPR serves as a promising approach for CA patients or refractory CA patients for whom ROSC is hard to achieve by CCPR, there are great difficulties in installing ECPR devices for the short window of resuscitation time under the critical circumstances of CA patients. Moreover, ECPR also requires qualified technical support and may be accompanied by serious complications such as bleeding, infection, and lower limb ischemia [15, 16]. In addition, the criteria for the appropriate population for ECPR treatment are not yet clear. Blindly using ECPR intervention may expose patients for whom ECPR is not necessary to high-risk invasive procedures and may lead to severe complications and causes financial burdens. Conversely, for patients who are suitable for ECPR, postponing ECPR treatment other than using ECPR timely may not only reduce its potential benefits but also increase additional risks. In summary, compared to CCPR, the implementation of ECPR has more uncertainties in population selection, technical operation, and timing. Therefore, there is still controversy over whether ECPR can improve the survival rate and neurological outcome of CA patients ultimately.

A meta-analysis by Kim et al. [17] included 10 cohort studies comparing the impact on the prognosis of ECPR versus CCPR in adult CA patients. The results showed that a higher survival rate and better neurological status appeared in patients who received ECPR. However, this meta-analysis may be seriously biased, for the original study included only involved cohort studies but with no randomized controlled trials (RCTs). Scquizzato et al. [18] conducted a meta-analysis which only included six studies to compare the prognosis of CA patients between ECPR group and CCPR group. The results showed that the survival rate and neurological prognosis of the ECPR group were better than those of the CCPR group, but the overall number of included original studies and sample size were considerably limited, which resulted in less convincing conclusions. In addition, the latest three RCTs assessing the efficacy of ECPR versus CCPR in the treatment of OHCA have yielded inconsistent conclusions. The RCT of Yannopoulos et al. [19] showed that compared with CCPR, ECPR significantly improved the survival rate and good neurological status, but the other two RCTs [20, 21] indicated that ECPR had no effect on survival rate and good neurological status. Based on these conclusions and newly published high-quality cohort studies with propensity-matched data, we believe that this meta-analysis can provide the latest insights into the difference of ECPR and CCPR on the prognosis of CA patients.

Methods

Protocol and registration

The protocol of this study was preregistered on PROSPERO (CRD42022385210). We conducted this meta-analysis based on the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) reporting guidelines.

Search strategy

All RCTs or cohort studies which were published by PubMed, EMBASE, and Cochrane Library from the inception to July 2023 were retrieved systematically by two reviewers. The following keywords or medical subject headings (MeSH) terms were used for retrieval: Extracorporeal cardiopulmonary resuscitation, Conventional cardiopulmonary resuscitation, heart arrest, resuscitation, Cardiopulmonary Bypass, and Extracorporeal Circulation. All search records were imported into ENDNOTE and duplicate documents were removed. They independently screened titles and abstracts for initial eligibility, as well as full texts for final eligibility. Any resulting discrepancies were settled by discussion with a third investigator.

Inclusion and exclusion criteria

We included the following studies: (1) The published clinical studies that included patients over 16 years of age with IHCA or OHCA; (2) Studies comparing the treatment of conventional CPR including manual CPR and & or mechanical CPR, with ECPR, including extracorporeal membrane oxygenation or cardiopulmonary bypass in CA patients; (3) Studies reporting the survival rate and neurological outcome after arrest at any time interval; (4) RCTs or observational studies published in English. Studies that include the following criteria are excluded: (1) Patients younger than 16 years old or pregnant; (2) CA Patients who did not receive CCPR or ECPR; (3) CA of non-medical causes such as trauma, drowning and poisoning; (4) letters, conference abstracts, case reports and reviews.

Data extraction

The two reviewers independently used a standard data extraction form to extract data from each study. When any disagreement occurs, it will be resolved through discussions with a third investigator. We collected the following information: (1) authors, publication year, study types, location of arrest, and countries and regions covered in the study; (2) sample size, number of patients with reported outcomes at any time interval after arrest, baseline characteristics of CA patients.

Definition of outcome indicators

The primary outcome was short-term (30 days after cardiac arrest) and long-term (≥ 90 days after cardiac arrest) survival with favorable neurological status. Secondary outcomes included post-arrest survival (measured as survival during follow-up at hospital discharge/1 month, 3–6 months, and 1 year). For long-term data, the longest available follow-up was used. We defined 1 or 2 points of a cerebral performance category (CPC) score [22] or a modified Glasgow Outcome Scale (MGOS >  = 4) [23] as good neurological outcome.

Quality assessment

The quality of the literature included was assessed by two researchers independently. Any unresolved disagreements between reviewers were resolved through discussion. We estimated the quality of RCTs based on the Cochrane Collaboration RCT risk assessment tool, the tool assessed the biases of 7 entries such as random sequence generation, allocation concealment, etc. We used the Newcastle–Ottawa Scale to appraise the quality of cohort studies. The scale assessed risk of bias in the cohort selection, group comparability, and outcomes, with a total score of 9 points.

Statistical analysis

We used the Review Manager version 5.3 for data analysis. The date that belongs to dichotomous outcomes (survival rate, incidence rate of good neurologic status) were summarized according to the Mantel Haenszel method, and we used the relative risk (RR) and 95% confidence interval (95% CI) to compare the results. All analyses were performed using the random-effects models, and heterogeneity was assessed by Q-value tests and I2 tests (I2 > 50% or P < 0.05, indicating substantive heterogeneity).

In this meta-analysis, for the purpose of ensuring the accuracy of comparing the impact of ECPR and CCPR on the prognosis of CA patients in the cohort studies, we performed a statistical analysis based on the propensity score matching method. Propensity score matching is a method that balances observed covariates in the two treatment arms by matching the propensity score representing the probability of receiving ECPR therapy [24]. The Covariates such as age, sex, comorbidities, bystander CPR, witnessed CA, initial rhythms, duration of CPR, and therapeutic hypothermia were used for propensity score matching.

We conducted a subgroup analysis by the location where the arrest occurs (IHCA/OHCA). The results were represented by forest plots. In addition, we evaluated the potential publication bias by using a visual funnel chart, and publication bias was considered to exist when the funnel is asymmetry [25]. The Begg's test and Egger’s test were used to evaluate the plot asymmetry. All significance tests were two-tailed, with P < 0.05 being the statistically significant difference.

Results

Retrieval results

Two independent researchers initially retrieved 6993 articles from three databases (PubMed: 1902, EMBASE: 4960, Cochrane: 131) and excluded 1342duplicate articles. By screening the titles and abstracts, we excluded 5601 articles, while 50 full-text articles were assessed as qualified. Three RCTs including 420 participants [19, 20, 26] and 14 cohort studies including 167,308 participants [7, 8, 27,28,29,30,31,32,33,34,35,36,37,38] were included eventually after reading the full text (Fig. 1).

Fig. 1
figure 1

Retrieval process of included studies

Full size image

Study characteristics

The meta-analysis included three RCTs and 14 cohort studies, nine of which had propensity-matched data. Among these 14 cohort studies, five were prospective cohort studies, and the other nine were retrospective cohort studies. The countries of these studies were in Asia (n = 11) and Europe (n = 6). In this meta-analysis, a total of 2308 patients received ECPR treatment and 165,420 patients received CCPR treatment. Among these studies, only three studies included both patients with IHCA and OHCA. Overall, the patients treated with ECPR were significantly younger than those treated with CCPR, and compared with the CCPR group, the ECPR group had more bystander CPR and patients with shockable rhythms, but the CPR duration was significantly prolonged. The basic characteristics of the included studies were summarized in Table 1.

Table 1 Basic characteristics of the study included in the meta-analysis
Full size table

Quality of studies

We used the Cochrane Collaboration's risk assessment tool to assess the quality of the three RCTs, no study was estimated having a “high risk of bias” in each domain. The results were illustrated in Table 2. The Newcastle Ottawa Scale was used to assess the bias risk of 10 cohort studies. The results showed that most studies were considered medium quality (Table 3).

Table 2 Risk assessment table of RCTs
Full size table
Table 3 Quality evaluation of cohort study
Full size table

Primary outcome

Short-term favorable neurological status

A total of eleven studies compared the short-term favorable neurological status between the ECPR group and the CCPR group, including 1442 patients receiving ECPR treatment and 51,221 receiving CCPR treatment. We found that ECPR improved the short-term neurological prognosis of CA patients (RR 2.88; 95% CI 1.96–4.23; p < 0.0001; I2 = 76%) (Fig. 2a). In the analysis with matched data, there were three RCTs and seven cohort studies with propensity-matched data. The results showed that a better neurological prognosis occurred in the ECPR group (RR 1.67; 95% CI 1.16–2.40; p = 0.005; I2 = 51%) (Fig. 2b).

Fig. 2
figure 2

A Short-term Neurological outcomes in accordance with crude data. B Short-term Neurological outcomes according to propensity-matched data

Full size image

Long-term favorable neurological status

Eleven studies that included 896 cases in the ECPR group and 1977 cases in the CCPR group reported the long-term favorable neurological status. We found that the long-term favorable neurological status of the ECPR group was markedly higher than the CCPR group (RR 2.11; 95% CI 1.40–3.19; p = 0.0004; I2 = 69%) (Fig. 3a). By analyzing the matched data, there were three RCTs and six cohort studies with propensity-matched data. As with the above results, the long-term favorable neurological status of the ECPR group tended to be higher than that of the CCPR group (RR 1.83; 95% CI 1.32–2.53; p = 0.0003; I2 = 14%) (Fig. 3b).

Fig. 3
figure 3

A Long-term neurological status in accordance with crude data. B Long-term neurological status according to propensity-matched data

Full size image

Secondary outcomes

Survival rate at various times of follow-up

Propensity-matched data from cohort studies were used to analyze secondary outcomes. There were three RCTs and nine cohort studies with propensity-matched data. The results indicated that the overall survival rate of the ECPR group was higher than that of the CCPR group (RR 1.51; 95% CI 1.20–1.89; p = 0.0004; I2 = 62%), especially the 3–6-month survival rate. (at discharge: RR 1.25, 95% CI 1.00–1.56, p = 0.05, I2 = 57%; at 3-6 months: RR 2.73, 95% CI 1.67–4.48, p < 0.0001, I2 = 0%; at one year: RR 1.92, 95% CI 1.14–3.25, p = 0.01, I2 = 0%) (Fig. 4).

Fig. 4
figure 4

Survival rate at various times of follow-up

Full size image

Subgroup analysis

Short-term and long-term favorable neurological status stratified by IHCA and OHCA

Ten studies reported the short-term favorable neurological status and eight studies reported the long-term favorable neurological status. The results showed that ECPR had different effects on favorable neurological status in patients with OHCA (short-term: RR 1.50, 95%CI 0.98- 2.29, I2 = 55%; long-term: RR 1.95, 95% CI 1.06–3.59, I2 = 11%). However, ECPR had significantly better effects than CCPR in patients with IHCA. (short-term: RR 2.18, 95%CI 1.24- 3.81, I2 = 9%; long-term: RR 2.17, 95% CI 1.19–3.94, I2 = 0%) (Table 4).

Table 4 Subgroup analysis
Full size table

Short-term and long-term survival rates stratified by IHCA and OHCA

Twelve studies reported short-term survival rate and six studies reported long-term survival rate. We found that the short-term and long-term survival rates of the ECPR group tended to be higher than that of the CCPR group in patients with IHCA (short-term: RR 2.03, 95%CI 1.30- 3.18, I2 = 0%; long-term: RR 1.92, 95% CI 1.14–3.25, I2 = 0%). However, ECPR had different effects on survival rates in patients with OHCA (short-term: RR 1.10, 95%CI 0.91- 1.34, I2 = 44%; long-term: RR 3.16, 95% CI 1.36–7.38, I2 = 0%) (Table 4).

Publication bias

The Begg's test and Egger's test were used to evaluate the publication bias of survival rate and good neurological prognosis at discharge. As shown in the following results, no publication bias was found (P > 0.05). [survival rate: Begg's test value (P = 1.000) and Egger's test value (P = 0.284), good neurological prognosis: Begg's test value (P = 0.436) and Egger's test value (P = 0.626)].

Discussion

Based on the RCTs and large sample size cohort studies latest published, we re-performed the meta-analysis to investigate whether ECPR improved the prognosis of CA patients compared to CCPR. In this meta-analysis, we found that ECPR significantly improved the short-term and long-term neurological outcomes and survival rate in CA patients. Although this result did accord with the meta-analysis by Scquizzato et al. [18] in 2022, our current meta-analysis included more original studies with larger sample sizes with more convincing findings.

Since this meta-analysis included more cohort studies, confounding factors may not be excluded completely. These confounders also made the basic characteristics of patients in the ECPR and CCPR groups differ significantly: patients treated with ECPR were younger and had a higher proportion of bystander CPR and shockable rhythms than those treated with CCPR, and these basic characteristics play a role in whether CA patients have a good prognosis [39, 40]. After adjusting the baseline characteristics of the cohort studies by the “propensity score matching method”, the results still suggested that patients treated with ECPR had a better neurological outcome than those treated with CCPR. The possible reasons are as follows:

Firstly, technically speaking, ECPR provides more oxygenated blood flow to CA patients, and this volume of oxygenated blood does not decrease over time. Due to the fact that CA patients can only tolerate short-term blood circulation disorders, the chance of survival will decline rapidly if CPR continues for more than 15–30 min [41]. Even high-quality chest compressions can only produce up to 25% of normal cardiac output, and the blood flow reduces with the prolongation of CPR duration [2]. Clearly, we should be committed to shortening the duration of cardiopulmonary resuscitation to minimize the risk of extensive hypoxic brain injury. Although the general information table in this meta-analysis showed that in most studies, the duration of cardiopulmonary resuscitation in the ECPR group was longer than that in the CCPR group, the prognosis of patients in the ECPR group was better than that in the CCPR group. This may be because ECPR can maintain blood and oxygen supply for CA patients through blood pumps and oxygenators, ensuring sufficient blood supply to important organs such as the brain and myocardium, thereby improving the survival rate and neurological prognosis [17, 42].

Secondly, ECPR may act as a bridge to subsequent invasive or appropriate therapies (e.g., percutaneous coronary intervention (PCI) or coronary artery bypass grafting (CABG)) for reversible causes of CA such as myocardial infarction and coronary ischemia, which is virtually infeasible in CCPR [43]. In addition, ECPR makes it easier to perform early therapeutic hypothermia (TH) that alleviates ischemia–reperfusion injury [29]. In our meta-analysis, we found that patients in the ECPR group were more likely to receive TH treatment than the CCPR group, which mainly because ECPR provided a basis for subsequent TH for CA patients. For another, hemodynamic instability in patients occurs more often in the CCPR group, and TH may aggravate the patient's condition [7]. Therefore, TH is less performed on the patients in the CCPR group.

In the secondary outcome, we found that the long-term survival rate of patients in the ECPR group was significantly higher than that of patients in the CCPR group. This result indicated that the benefits of ECPR for patients with CA may be long-term and related to survival with good neurological outcome. Furthermore, the improvement of long-term survival rate in CA patients may be attributed to the recovery of some brain functions.

We conducted a subgroup analysis on the survival rate and favorable neurological status of IHCA and OHCA. Firstly, we found that patients receiving ECPR treatment had better long-term neurological outcomes than those receiving CCPR treatment in both IHCA and OHCA. ECPR provides stable systemic perfusion and rapid pathways for PCI or CABG, enabling it to treat potential causes of cardiac arrest [42, 44]. In addition, ECPR makes it easier to perform TH in CA patients, thereby reducing oxygen consumption and alleviating brain edema, which improves the recovery of nervous system function [45].

Secondly, the results suggested that the patients with IHCA had higher short-term and long-term survival rate in the ECPR group, which was similar to the meta-analysis results of Gravesteijn et al. [46]. However, in terms of short-term survival rate of OHCA patients, we found that there was no obvious difference between the ECPR group and the CCPR group, the cause of which may be that OHCA patients suffer from a longer period of no-flow or low-flow circulation status, let alone the varied and complex potential causes of CA.

Limitations

The limitations of our meta-analysis are as follows. Firstly, there are only three RCTs but 14 cohort studies included in this meta-analysis. Although we try to use high-quality research with propensity score matching method to minimize the impact of confounding factors, the influence of unknown confounding factors in cohort studies and the limitation of insufficient RCTs should also be put into consideration. Secondly, among the indications for ECPR, the inclusion criteria vary across study populations in terms of the location of the arrest, duration of no-flow and CPR. Moreover, most studies on ECPR are observational, and ECMO usage cannot be blinded. Consequently, the subjective selection of ECPR participants may introduce bias into the estimation of survival and neurological outcomes. Finally, since the included studies come from different countries and regions, it inevitably leads to varying heterogeneity between studies for differences in the quality of ECPR teams, the characteristics of CA patients, medical facilities, emergency medical service systems, the systematic accessibility to ECPR.

Conclusions

Compared with CCPR, ECPR had significantly better effects on good neurological prognosis and survival rate. In addition, ECPR has the potential to improve neurological status and survival rate of IHCA patients, whereas its effect on the short-term outcome of patients with OHCA was not significant. More high-quality studies are still needed to investigate the potential benefits of ECPR in CA patients.